Sentiment and the Securities and Exchange Commission (sec) Explored

This article explores research which highlights the consequences of the work of Loughren and McDonald. (Tim Loughran and Bill McDonald, 2016, Textual Analysis in Accounting and Finance: A Survey, Journal of Accounting Research, 54:4,1187-1230.)

Loughren and McDonald identified lists of words from SEC filings which they felt indicated various sentiments or emotions, such as: positive, negative, litigious, constraining, interesting, and uncertainty.

One consequence has been that investment firms began automatically monitoring SEC filings for problematic words from these lists, perhaps as a first filter, or perhaps instead of understanding the true situation of the company.

As an inevitable consequence of human nature, businesses that make SEC filings or other reports responded by censoring their written language to avoid certain words from the Loughren and McDonald dictionaries, thus potentially negating their utility.

This sort of behaviour shows very clearly that language is a living product of human cognition, and changes with time, knowledge, and need. There is never just one fixed meaning for a word. Leximancer adaptive text modelling understands this, and so is much less vulnerable to people manipulating their words.

As a powerful and convenient way of checking or monitoring these sentiment dimensions within corporate documents, Leximancer has made available a dictionary of the sentiment word lists from Loughren and McDonald (as updated in 2019). These can be simply loaded into Leximancer as "User Defined Tags" for classifying, indexing, and exploring collections of company documents using these dimensions. Email us for a copy.